[Clang][AVX512][BUILTIN] Adding intrinsics for compressstore{df|di|sf|si} instruction set.

include/clang/Basic/BuiltinsX86.def

 TARGET_BUILTIN(__builtin_ia32_expandsi512_mask, "V16iV16iV16iUs","","avx512f")
 TARGET_BUILTIN(__builtin_ia32_cvtps2pd512_mask, "V8dV8fV8dUcIi","","avx512f")
 TARGET_BUILTIN(__builtin_ia32_movapd512_mask, "V8dV8dV8dUc","","avx512f")
 TARGET_BUILTIN(__builtin_ia32_movaps512_mask, "V16fV16fV16fUs","","avx512f")
 TARGET_BUILTIN(__builtin_ia32_movapd128_mask, "V2dV2dV2dUc","","avx512vl")
 TARGET_BUILTIN(__builtin_ia32_movapd256_mask, "V4dV4dV4dUc","","avx512vl")
 TARGET_BUILTIN(__builtin_ia32_movaps128_mask, "V4fV4fV4fUc","","avx512vl")
 TARGET_BUILTIN(__builtin_ia32_movaps256_mask, "V8fV8fV8fUc","","avx512vl")
+TARGET_BUILTIN(__builtin_ia32_compressstoredf512_mask, "vV8d*V8dUc","","avx512f")
+TARGET_BUILTIN(__builtin_ia32_compressstoredi512_mask, "vV8LLi*V8LLiUc","","avx512f")
+TARGET_BUILTIN(__builtin_ia32_compressstoresf512_mask, "vV16f*V16fUs","","avx512f")
+TARGET_BUILTIN(__builtin_ia32_compressstoresi512_mask, "vV16i*V16iUs","","avx512f")
 
 #undef BUILTIN
 #undef TARGET_BUILTIN

lib/Headers/avx512fintrin.h

 _mm512_maskz_mov_ps (__mmask16 __U, __m512 __A)
 {
   return (__m512) __builtin_ia32_movaps512_mask ((__v16sf) __A,
              (__v16sf)
              _mm512_setzero_ps (),
              (__mmask16) __U);
 }
 
+static __inline__ void __DEFAULT_FN_ATTRS
+_mm512_mask_compressstoreu_pd (void *__P, __mmask8 __U, __m512d __A)
+{
+  __builtin_ia32_compressstoredf512_mask ((__v8df *) __P, (__v8df) __A,
+            (__mmask8) __U);
+}
+
+static __inline__ void __DEFAULT_FN_ATTRS
+_mm512_mask_compressstoreu_epi64 (void *__P, __mmask8 __U, __m512i __A)
+{
+  __builtin_ia32_compressstoredi512_mask ((__v8di *) __P, (__v8di) __A,
+            (__mmask8) __U);
+}
+
+static __inline__ void __DEFAULT_FN_ATTRS
+_mm512_mask_compressstoreu_ps (void *__P, __mmask16 __U, __m512 __A)
+{
+  __builtin_ia32_compressstoresf512_mask ((__v16sf *) __P, (__v16sf) __A,
+            (__mmask16) __U);
+}
+
+static __inline__ void __DEFAULT_FN_ATTRS
+_mm512_mask_compressstoreu_epi32 (void *__P, __mmask16 __U, __m512i __A)
+{
+  __builtin_ia32_compressstoresi512_mask ((__v16si *) __P, (__v16si) __A,
+            (__mmask16) __U);
+}
 
 #undef __DEFAULT_FN_ATTRS
 
 #endif // __AVX512FINTRIN_H

test/CodeGen/avx512f-builtins.c

__m512 test_mm512_mask_mov_ps(__m512 __W, __mmask16 __U, __m512 __A) {
   return _mm512_mask_mov_ps(__W, __U, __A); 
 }
 
 __m512 test_mm512_maskz_mov_ps(__mmask16 __U, __m512 __A) {
   // CHECK-LABEL: @test_mm512_maskz_mov_ps
   // CHECK: @llvm.x86.avx512.mask.mova.ps.512
   return _mm512_maskz_mov_ps(__U, __A); 
 }
+
+void test_mm512_mask_compressstoreu_pd(void *__P, __mmask8 __U, __m512d __A) {
+  // CHECK-LABEL: @test_mm512_mask_compressstoreu_pd
+  // CHECK: @llvm.x86.avx512.mask.compress.store.pd.512
+  return _mm512_mask_compressstoreu_pd(__P, __U, __A); 
+}
+
+void test_mm512_mask_compressstoreu_epi64(void *__P, __mmask8 __U, __m512i __A) {
+  // CHECK-LABEL: @test_mm512_mask_compressstoreu_epi64
+  // CHECK: @llvm.x86.avx512.mask.compress.store.q.512
+  return _mm512_mask_compressstoreu_epi64(__P, __U, __A); 
+}
+
+void test_mm512_mask_compressstoreu_ps(void *__P, __mmask16 __U, __m512 __A) {
+  // CHECK-LABEL: @test_mm512_mask_compressstoreu_ps
+  // CHECK: @llvm.x86.avx512.mask.compress.store.ps.512
+  return _mm512_mask_compressstoreu_ps(__P, __U, __A); 
+}
+
+void test_mm512_mask_compressstoreu_epi32(void *__P, __mmask16 __U, __m512i __A) {
+  // CHECK-LABEL: @test_mm512_mask_compressstoreu_epi32
+  // CHECK: @llvm.x86.avx512.mask.compress.store.d.512
+  return _mm512_mask_compressstoreu_epi32(__P, __U, __A); 
+}